OpenAI Debuts Custom 'Jalapeño' AI Chip with Broadcom to Boost LLM Inference

OpenAI, the developer behind popular AI models like ChatGPT, has officially introduced its inaugural custom artificial intelligence chip, named 'Jalapeño.' This groundbreaking hardware was co-developed with semiconductor giant Broadcom and is meticulously designed to optimize AI inference workloads, which are crucial for running large language models (LLMs) efficiently. The introduction of Jalapeño underscores OpenAI's commitment to bolstering its AI infrastructure and gaining greater control over its computing resources. The company aims to mitigate the high costs and supply constraints associated with acquiring powerful GPUs from third-party vendors, particularly Nvidia, which currently dominates the AI hardware market. By developing in-house silicon, OpenAI seeks to achieve better performance-per-watt ratios and tailor hardware precisely to the demands of its evolving LLMs. According to OpenAI hardware chief Richard Ho, the Jalapeño processor is built to perform speedily and efficiently with current and future iterations of LLMs. Broadcom CEO Hock Tan reportedly stated that the chip's capabilities are comparable to Nvidia's Blackwell chips and Google's Tensor Processing Units (TPUs). The collaboration saw OpenAI's engineers design the chip, with Broadcom contributing its expertise in custom silicon manufacturing. The design process for Jalapeño was remarkably swift, taking approximately nine months, partly due to the use of AI to accelerate specific design tasks. OpenAI intends to deploy the Jalapeño chips in its data centers starting by the end of this year. This initiative represents the initial phase of a broader, multi-generational chip development plan, signaling a long-term strategy to integrate custom hardware deeply into its AI ecosystem. The company has already reported that initial tests show the chip operating at target power and performance levels with its GPT-5.3-Codex-Spark AI model. This strategic pivot into custom AI hardware is expected to not only boost OpenAI's revenue growth but also potentially increase investor interest ahead of any future public offering, especially as competitors like Anthropic also explore similar avenues.